1
Cryptocurrency Price Analytics Report
To what extent can you use descriptive, diagnostic and predictive analytics
to predict future cryptocurrency price trends?
A Business & Data Analytics report investigating factors influencing cryptocurrency and
blockchain technology adoption.
Felix McAuliffe
MSc Business Analytics
2
Table of Contents
Introduction to Blockchain: The Trust Machine.................................................................. 3
Descriptive Statistical Analysis of Five Cryptocurrencies .................................................... 4
Crypto price charts over the last 4 years ................................................................................... 4
Cumulative returns of 5 Cryptos ............................................................................................... 5
Cumulative returns (Adjusted for comparability) ...................................................................... 5
Comparison of Daily Volatilities ................................................................................................ 6
Comparison of Return Distributions.......................................................................................... 7
Comparison of boxplots with and without outliers .................................................................... 7
Correlation Between Cryptocurrencies ..................................................................................... 8
Diagnostic Analysis of Cryptocurrencies ............................................................................ 9
Ripple (XRP) ............................................................................................................................. 9
Predictive Analysis of Cryptocurrency Prices .................................................................... 10
Conclusion: Going Above & Beyond ................................................................................ 13
Reference: ...................................................................................................................... 14
Appendix: ....................................................................................................................... 16
3
Introduction to Blockchain: The Trust Machine
An interesting story to turn to when initially familiarising oneself with cryptocurrencies and
blockchain technology is that of William Henry Furness III, an American anthropologist who
travelled to a miniscule island in the Pacific Ocean called Yap (‘The Forbidden Island’) in
1903. On this secluded island is a monetary system like no other on earth. Their culture thrives
on exchanging the world’s heaviest and physically largest form of currency, the “rai” or “fei”,
which is essentially chunks of limestone quarried in Palau some 400 kilometres away and rafted
over to Yap. The difficulty of importing these stones as well as the currency’s inherent scarcity
contribute to its capacity to maintain value within the island’s economy. Perhaps the most
remarkable feat of Fei is that the stone doesn’t even have to be moved in order for someone to
spend it. Rather, the ownership of each stone is logged in Yap’s collective accounting ledger:
the memory of its residents. The extraordinary nature of Yap’s financial system has intrigued
economists such as Milton Friedman for decades, comparing it to the gold standard, citing that
as long as people have faith in the value of something, anything can be the trusted and
transacted monetary device. Philosophically, community consensus of who owns the Fei is the
true ‘money’, rather than the physical stone itself.
The global financial crisis in 2008 was an inflection point in regard to whether governments
and central banks should really be trusted to regulate our economy. Amidst financial
catastrophe, the pseudonymous individual(s) known as Satoshi Nakamoto distributed their
whitepaper called “Bitcoin: A Peer-to-Peer Electronic Cash System”. A prominent theme in
the paper is as follows: “The root problem with conventional currency is all the trust that’s
required to make it work. The central bank must be trusted not to debase the currency, but the
history of fiat currencies is full of breaches of that trust”. In a way, Bitcoin is a duplicate of
Yap’s stone technology. When Bitcoin is sent from one address to another, there is no physical
exchange of ownership over the digital tokens. Equally, when a 4-ton Fei stone is used to pay
for a wedding, the stone does not move – just its ownership. In the island, trust on this level
works because the community is tightly-knit, however, who’s to say the same singular Bitcoin
can’t be sent to multiple people in a copy-and-paste fashion? That comes down to Satoshi’s
innovative ‘blockchain’.
The Bitcoin blockchain is the pioneer of blockchains. The blockchain is an immutable database
of ‘blocks’ linked together, where each block houses a collection of transactions from a
recorded period. Once a block is full, a new one is added to the chain, hence, ‘blockchain’.
Blocks cannot be amended, only appended, meaning that every confirmed transaction is final
and recent transactions are simply added on once verified by nodes on the network. Bitcoin is
an opensource public ledger where people can transfer value among Bitcoin wallets. These
transfers are proven mathematically by ‘miners’ which are computers optimised to confirm
pending transactions within the distributed consensus system. Miners are rewarded with either
newly minted coins accompanying new validated blocks or transaction fees from transactions
in the block for their computational hashing power (verifying transactions), but only if they’re
the first to discover the solution to the complex, cryptographic hashing puzzle.
With all considered, it is commonly understood that blockchains are ‘trustless’ ‘confidence
machines’ that do not require the presence of a financial intermediary (banks, PayPal) to certify
the flow of transactions. Rather, nodes perform complex calculations in a competitive race to
verify transactions and get compensated for it, followed by a consensus mechanism that ensures
validators’ calculations are accurate and honest before the block is substantiated and added to
the blockchain.
4
Nevertheless, despite blockchains being relentless trust machines, there is significant resistance
to their adoption due to various political, social, technological and regulatory factors. First, a
price analysis of five cryptocurrencies is conducted in an exploratory manner. No prior
expectations were set before setting forth in this analysis, instead, it is employed to uncover
trends and insights into the primary factors influencing adoption and use of these five
cryptocurrencies, represented by price movement over time.
Descriptive Statistical Analysis of Five Cryptocurrencies
Crypto price charts over the last 4 years
What we see here are the price charts of five cryptocurrencies (cryptos) over the last 4 years.
Bitcoin (BTC), Ethereum (ETH), Ripple Labs (XRP), DogeCoin (DOGE), and Zilliqa (ZIL)
were chosen for analysis for various reasons. BTC, ETH, XRP & DOGE have all been among
the top cryptocurrencies with the highest market capitalisation at one point or another over the
past few years, making them some of the most important names in this relatively new asset
class. Whilst ‘Stablecoins’ have also circulated among the top most valued cryptos in terms of
market capitalisation, they will not be heavily focused on in this report as their unit prices
cannot be influenced. The “crypto bubble” burst in 2017 after all-time-highs (ATH) were
reached for various cryptos, however, this period will not be studied in depth as the crypto
environment has shifted significantly since that historical time frame.
At first glance, we notice that these five cryptos were in a serious recession period, in general,
from the beginning of 2019 until price traction again began to build at the beginning of 2020.
Also notable, we see that each crypto’s price scale is independent of one another. For example,
BTC’s Y-axis grows upwards of US$60,000, whereas ETH’s Y-axis price scale only ascends
until just below US$5,000. For this reason, a cumulative price standardisation process is
required to make the cryptos’ price movement more comparable.
5
Cumulative returns of 5 Cryptos
The daily relative price changes, or percent changes, illustrate the rate of cumulative growth
over time for the different cryptocurrencies. Interestingly, the relative cumulative growth of
DOGE far exceeds that of the other four cryptos. Below is a chart without DOGE to make
better visual comparisons.
Cumulative returns (Adjusted for comparability)
Now it is possible to recognise the price trends relative to one another more accurately. As
shown, BTC out-performed the other three cryptos and acted as the asset class ‘market-maker’
in the sense that the smaller market cap cryptos only began to appreciate once Bitcoin led the
way. ETH’s macro price trend seems to mirror that of BTC’s, though, initially lagging before
catching up toward the end of 2021. In summary, we need cumulative returns to make different
assets comparable.
6
Comparison of Daily Volatilities
As shown above, we see the volatilities of cryptocurrencies on daily fluctuations. The most
notable fluctuation is around mid-May 2020, where we observe a stark 35-40% decrease in
value across all five assets. Somewhat surprisingly, DOGE was the least affected, ‘only’ losing
around 30% of it’s value. XRP and DOGE demonstrate the highest volatility from the
beginning of 2021 onwards. In general, the volatility seems to be least in BTC’s case and most
in ZIL’s case.
7
Comparison of Return Distributions
As evident, on the x-axis, we see the returns; positive on the right-hand side, negative on the
left. On the y-axis, we see the frequency, or number of occurrences, of various daily returns
over the last 4 years. It’s very interesting to compare the percentage values of daily returns in
this way. In Bitcoin and DogeCoin’s case, it is evident that the most frequent daily percent
change is actually a positive one since the highest point on the distribution curve sits slightly
to the right of zero. In ZIL’s case, we observe a much flatter shape compared to the other four
cryptos, where the tails on either side (particularly on the left) are larger than that of the other
cryptos.
Comparison of boxplots with and without outliers
In the first boxplot, we see the distribution of daily returns with outliers, the largest and most
significant outlier being attributed to DOGE, where in one occasion there was a return of
approximately 270%. The number of outliers is so high that the boxes are compressed, and we
can’t actually gain any insights from the first visual representation. For this reason, a boxplot
graph without outliers was also necessary in order to meaningfully visualise the data and gain
valuable insights such as the interquartile range (blue box) and median value (green line) of
each crypto.
8
At first glance, it appears as though the
medians (midpoint of frequency
distribution) are equal to zero, however,
upon closed inspection it’s clear that
BTC, ETH and ZIL have median values
of slightly above zero, while XRP and
DOGE sit just below zero. Here, the
interquartile range appears to be the
highest for Zilliqa, followed by
Ethereum, Ripple, DogeCoin, and then
Bitcoin.
Correlation Between
Cryptocurrencies
Let’s have a look at how the cryptos are
correlated against each other. On top,
there’s the correlation of the five
cryptocurrencies and at the bottom there’s
the visualisation of the correlation matrix.
The first row details the correlation between
Bitcoin and the other cryptos, and what we
notice is that there is quite a high correlation
for ETH and ZIL against BTC and very low
correlation for DOGE. This is very
interesting because when we look at
DOGE’s correlation against the other
cryptos we can see that, in general, DOGE
has very low correlation across the board,
whereas when we compare ETH to the other
cryptos, there is generally quite high
correlation.
9
Diagnostic Analysis of Cryptocurrencies
The previous section was concerned with a descriptive statistics analysis (technical analysis)
approach to price analysis and prediction. This section explores a diagnostic analysis
(fundamental analysis) of the five cryptocurrencies which is exemplified by real-world factors
influencing the intrinsic value of a crypto’s ecosystem, judging by its use-cases and subsequent
adoption or abandonment.
This fundamental analysis attempts to link events to ensuing price movement, where the price
movement occurs as a result of new specific use-cases for crypto tokens, news pertaining to
potential future adoption or rejection of the token, and various other Political, Economic,
Societal, Technological, Legal, or Environmental contributing factors.
Ripple (XRP)
1. Ripple onboards 10 more companies to its roster (Economic)
Ripple Labs successfully negotiated terms with 10 financial institutions, leading to an
immediate spike in network use, and thus, positive price movement thereafter. At the time,
Ripple was one of the only blockchain-led institutions offering commercial products and
services to its clients.
2. American Express Partnership (Economic)
International payments settlement can take days with a bank, which is why the deployment of
FX International Payments (FXIP) was such a ground-breaking, value-added initiative by
AMEX and Ripple. B2B payments simply were not being verified quickly enough, prompting
AMEX to search for more innovative blockchain-powered solutions with Ripple.
3. XRP listed on 50 exchanges, breaks $1 (Societal)
At the beginning of 2017, XRP was only available to trade on six exchanges. By the end of
2018, XRP was trading on 50 exchanges, partly due to the successful roll out of xRapid. Being
listed on so many exchanges made XRP multi-fold more accessible for retail investors to
participate in the trade of the digital asset.
4. Ripple vs R3 reach court settlement (Legal)
10
For an entire year, Ripple and R3 battled in court over Ripple’s failure to respect and follow
through with a standing purchase agreement, which would have allowed R3 permission to
purchase five billion XRP at $0.00085 even though the trading price per XRP was $0.26.
Understandably, investors may have been hesitant to purchase XRP or partner with Ripple
amid this period since, had R3 won the case, the rate of each XRP would have plummeted.
Investors celebrated that day, as reflected by a 113% price increase mere hours after the
settlement.
5. SEC vs XRP begins (Legal)
The Securities Exchange Commission (SEC) prepared to take legal action against Ripple for
the ‘sales of unregistered securities’. The case persists to this day and is widely regarded as a
huge over-step of authority on the SEC’s behalf. After a slight dip in price, XRP’s price prevails,
potentially due to the falsifiable claims made by the SEC. For instance, the SEC claims the
Howey test proves Ripple to be in violation of distributing unregulated securities but
continuously fails to justify how the Howey test applies to cryptocurrencies. Additionally, it
appeared as though Jay Clayton, former SEC Chairman, made a hasty final decision in office
when litigating Ripple, the point of contention is, in 2018, Clayton decided Bitcoin did not
constitute a security under US law. Granted, Ripple and XRP operate in very different ways to
the Bitcoin network.
6. Self-branded “XRP Army” Unites (Societal)
Despite SEC pushing forward with their suit charging two Ripple executives for $1.38bn in
‘unregistered sales of securities’, CoinDesk’s “The Hash” speculate that the 91% price jump
over the first week of April, 2021, was the works of the online “XRP Army” community. They
even protested at the SEC HQ in Washington, DC.
Obviously, there are countless more news articles and conversations constantly on-going
regarding cryptocurrencies and blockchain-powered companies than is possible to keep track
of. When considering crypto price prediction, it appears evident that, as shown above, there is
some correlation between headlines posted on the internet and subsequent price movement. A
substantial majority of crypto and blockchain discourse takes place on social media platforms
which is presents a huge opportunity to gain historical as well as live public sentiment
pertaining to the research topic of choice. Mata et al. managed to show a positive correlation
between (positive/ negative sentiment) tweets mentioning Bitcoin and Bitcoin price trends.
Similarly, Stenqvist and Lönnö studied a dataset of over two million tweets discussing Bitcoin
over a one month period to explain Bitcoin price fluctuations, citing that machine learning
application and text classification techniques should be explored further to improve the
accuracy of such predictive models.
In light of researchers’ success in correlating text sentiment with cryptocurrency price
fluctuations and even showing promising steps toward building an accurate price prediction
model, this research paper deems it necessary to strive further in this important forecasting
field. While it is beyond the scope of this paper to prescribe any technical recommendations as
to exactly how a machine learning algorithms using natural language processing (NLP) should
be built, it is recommended that further research be conducted to develop distinctive sentiment
lexicon datasets for the specific use-case of price prediction.
Predictive Analysis of Cryptocurrency Prices
This report uses the timeseries prediction model developed by Facebook’s Data Science Team
(fbprophet). Firstly, two backtests of the model are conducted in retrodiction, one from a year
back and another from six months prior.
11
Two time horizons are required, “end1” for the 1 year backtest and “end2” for the 6 month
backtest, where “start1” indicates where the backtest begins from.
After that, two online requests are required for each time horizon, where “df1” is for the 1 year
backtest and “df2” is for the 6 month backtest.
On the left-hand side
are two prediction
models, where the
dataframes above
define the parameters
of the models. Both
models “modelfb1”
and “modelfb2” are fit
to dataframes “df1”
and “df2”, respectively.
Below, the future time
horizons are defined. The “make_future_dataframe” function of the model is used to define the
365 day backtest (“future”), and 180 days for the 6 month backtest (“future1”).
12
As described, we first have to backtest the data to determine whether this price prediction
model is accurate in predicting future price discovery. The degree to which this backtest proves
to reflect realised crypto price data of the last 6 months and 1 year supports analysts’ confidence
in its ability to accurately predict trends leading into the future.
As we can see after plotting the prediction of the 1 year and the 6 month backtest, with the aid
of the actual Bitcoin price chart (black box overlaying, adapted from Binance Cryptocurrency
Exchange platform), the model would have failed to predict the price accurately both times.
“modelfb1.plot” drastically under-predicted BTC’s price movement 1 year into the future,
whereas “modelfb2.plot” clearly over-predicted BTC’s price movement 6 months into the
future, in retrospect.
13
Finally, the third model, “BTC 6 months in the future”, is plotted using the original DataFrame
and a 6-month time horizon to predict Bitcoin’s price 180 days in the future.
Conclusion: Going Above & Beyond
This analytical report presents descriptive, diagnostic, and predictive analytics techniques for
the purpose of predicting future cryptocurrency prices. However, it is evident that Facebook’s
general spatiotemporal prediction algorithm (Prophet) is not optimised for, nor specialised
enough to be geared toward predicting Bitcoin’s future price trends – yet. To extend this
research, interested parties should consider incorporating varied data types with a multi-
dimensional data processing procedure. A supervised machine learning (ML) classification
model and unsupervised ML dimensionality reduction model can be fitted to data collected
using natural language processing techniques via API’s to retrieve such data from social media
platforms.
14
Reference:
American Express Introduces Blockchain-enabled, Cross-border Payments. (n.d.). Ripple. 2021,
from https://ripple.com/ripple_press/american-express-introduces-blockchain-enabled-cross-
border-payments/
Blockchain and the Island of Yap. (n.d.)., from https://rsmus.com/what-we-
do/services/blockchain-consulting/featured-topics/blockchain-basics/blockchain-and-the-
island-of-yap.html
Browne, R. (2018, October 1). Ripple’s cryptocurrency product goes live for the first time with
three financial firms. CNBC. https://www.cnbc.com/2018/10/01/ripple-xrp-cryptocurrency-
product-xrapid-goes-live-for-first-time.html
Godbole, O. (2021, April 6). XRP Rises Above $1 for First Time Since March 2018, Despite SEC
Shadow. https://www.coindesk.com/markets/2021/04/06/xrp-rises-above-1-for-first-time-
since-march-2018-despite-sec-shadow/
Goldstein, J., & Kestenbaum, D. (2010, December 10). The Island Of Stone Money. NPR.
https://www.npr.org/sections/money/2011/02/15/131934618/the-island-of-stone-money
Hash – The Puzzle of Bitcoin (What Are the Miner’s Mining?). (n.d.). VpnMentor., from
https://www.vpnmentor.com/blog/hash-puzzle-bitcoin/
Matta et al. - Bitcoin Spread Prediction Using Social And Web Sea.pdf. (n.d.)., from http://ceur-
ws.org/Vol-1388/DeCat2015-paper3.pdf
Matta, M., Lunesu, I., & Marchesi, M. (n.d.). Bitcoin Spread Prediction Using Social And Web
Search Media. 10.
Nakamoto, S. (2008). Bitcoin: A Peer-to-Peer Electronic Cash System. 11.
Nakamoto—Bitcoin A Peer-to-Peer Electronic Cash System.pdf. (n.d.). from
https://www.ussc.gov/sites/default/files/pdf/training/annual-national-training-
seminar/2018/Emerging_Tech_Bitcoin_Crypto.pdf
R3 Bought at Least $240 Million Worth XRP, SEC Lawsuit Shows. (2020, December 25).
Financial and Business News | Finance Magnates.
https://www.financemagnates.com/cryptocurrency/news/r3-bought-at-least-240-million-
worth-xrp-sec-lawsuit-shows/
Ripple and R3 Reach Settlement. (2020). Ripple., from https://ripple.com/ripple_press/ripple-and-
r3-reach-settlement/
SEC Chair Says Bitcoin Is Not A Security. (2018). Investopedia. from
https://www.investopedia.com/news/sec-chair-says-bitcoin-not-security/
SEC sues Ripple and two executives for selling XRP crypto: A $1.3 billion illegal security
offering—CNN. (). from https://edition.cnn.com/2020/12/23/investing/ripple-sec-lawsuit-
trnd/index.html
SEC.gov | SEC Charges Ripple and Two Executives with Conducting $1.3 Billion Unregistered
Securities Offering. (2021, Jan)., from https://www.sec.gov/news/press-release/2020-338
Stenqvist and Lönnö—Predicting Bitcoin price fluctuation with Twitter s.pdf. (2015)., from
http://www.diva-portal.org/smash/get/diva2:1110776/FULLTEXT01.pdf
Stenqvist, E., & Lönnö, J. (2015). Predicting Bitcoin price fluctuation with Twitter sentiment
analysis. 37.
Ten More Financial Institutions Join Ripple’s Global Payments Network. (2018). Ripple. 1, from
https://ripple.com/ripple_press/ten-financial-institutions-join-ripples-global-payments-
network/
XRP Available on 50 Exchanges. (2017, December 21). Ripple. https://ripple.com/insights/xrp-
now-available-on-50-exchanges-worldwide/
XRP purchasers back Ripple, arguing that it is not a security. (2020). Cointelegraph., from
https://cointelegraph.com/news/xrp-purchasers-back-ripple-arguing-that-it-is-not-a-security
Word count: 3218
15
